Home

Welcome to the 3DMapping wiki! Here you will find brief descriptions of each file in our repository. Click on a header, then read the notes below each file name.

To run our software, follow these instructions:

1. Generate Raw Data

• Start CARLA: ./CarlaUE4.sh

  • We tested on Carla V0.9.13 with Python 3.7. You can follow the official document on getting started with Carla.

• Edit parameters in the bottom of GatherCarlaData.py as desired.

  • storage_dir (where output are stored)
  • new_world (town number)
  • num_vehicles (total vehicles)
  • num_walkers (total pedestrians)
  • num_sensors (total lidars)

• Run GatherCarlaData.py to begin saving raw data.

• Stop CARLA when enough data has been obtained. The maps of town 06, 07, 10 could be downloaded to generate more data.

2. Process Data

• Edit parameters in the bottom of SceneCompletion.py as desired.

  • cylindrical (boolean for storing voxels in cylindrical or cartesian coordinates)
  • parent_dir (where output are stored)
  • t_start, t_end (starting frame and ending frame for processing)
  • free_res (distance in meters to ray trace free space)

• Run SceneCompletion.py to generate processed semantic scenes.

3. Train Model

• Edit the parameters at the top of train_carla.py with the location of your completed data, and the model you wish to train.
• Run train_carla.py, and it will train your model while saving the training meta data and weights.

4. Test Model

• Edit test_carla.py with your model location and boolean flags for any tests you wish to perform.
• Run test_carla.py to obtain your results.